added support for ignoring fields from xml document

RalphTro · Jun 8, 2024 · 812da70 · 812da70
1 parent b68b60f
commit 812da70
Show file tree

Hide file tree

Showing 7 changed files with 171 additions and 0 deletions.
diff --git a/epcis_event_hash_generator/xml_to_py.py b/epcis_event_hash_generator/xml_to_py.py
@@ -65,6 +65,7 @@
 
 import logging
 import xml.etree.ElementTree as ElementTree
+from lxml import etree
 from typing import Tuple
 
 _expansions = {"gs1:": "https://gs1.org/voc/", "cbv:": "https://ref.gs1.org/cbv/"}
@@ -168,13 +169,46 @@ def _xml_to_py(root, sort=True):
     return obj
 
 
+def remove_xml_declaration(xml_string):
+    """
+    Removes the <?xml> tag from the beginning of an XML string if present.
+    """
+    if xml_string.startswith("<?xml"):
+        # Find the end of the processing instruction
+        end_pos = xml_string.find("?>") + 2  # Include the ?> characters
+        return xml_string[end_pos:]
+    else:
+        return xml_string
+
+
+def get_ignore_field_prefix_ns(xmlStr: str):
+    """
+    if presents, gets all fields to be ignored from events of EPCIS document.
+    """
+    xml_str_without_decl = remove_xml_declaration(xmlStr)
+
+    all_namespaces = etree.fromstring(xml_str_without_decl).nsmap
+
+    ignore_field_ns_prefix = None
+    for key, value in all_namespaces.items():
+        if value == 'https://repository-x.example.com/':
+            return key
+
+    return ignore_field_ns_prefix
+
+
 def event_list_from_epcis_document_str(xmlStr: str) -> Tuple[str, str, list]:
     """
     Read EPCIS XML document and generate the event List in the form of a simple python object
     """
     try:
         data = _remove_extension_tags(xmlStr)
 
+        ignore_field_ns_prefix = get_ignore_field_prefix_ns(xmlStr)
+
+        if ignore_field_ns_prefix is not None:
+            data = data.replace(ignore_field_ns_prefix + ':', '')
+
         root = ElementTree.fromstring(data)
 
         eventList = root.find("*EventList")
@@ -184,6 +218,14 @@ def event_list_from_epcis_document_str(xmlStr: str) -> Tuple[str, str, list]:
 
         if not eventList:
             raise ValueError("No EventList found")
+
+        # remove all fields to be ignored
+        for field_to_remove in root.findall("ignoreFields/*"):
+            for event in eventList:
+                el = event.find(field_to_remove.tag)
+                if el is not None:
+                    event.remove(el)
+
     except (ValueError, OSError) as ex:
         logging.error(ex)
         logging.error("Input string does not contain a valid EPCIS XML document with EventList.")

diff --git a/tests/examples/epcisXmlDocHavingEventWithIgnoreFields.hashes b/tests/examples/epcisXmlDocHavingEventWithIgnoreFields.hashes
@@ -0,0 +1,2 @@
+ni:///sha-256;cec19d047cf3b05d02cf719d9c35bc468e911d7b9f5d415ba67a074a104c28d2?ver=CBV2.0
+ni:///sha-256;cec19d047cf3b05d02cf719d9c35bc468e911d7b9f5d415ba67a074a104c28d2?ver=CBV2.0
diff --git a/tests/examples/epcisXmlDocHavingEventWithIgnoreFields.prehashes b/tests/examples/epcisXmlDocHavingEventWithIgnoreFields.prehashes
@@ -0,0 +1,3 @@
+eventType=ObjectEventeventTime=2015-03-30T08:58:56.591ZeventTimeZoneOffset=+02:00quantityListquantityElementepcClass=https://id.gs1.org/01/04054739999612/10/2015-03-30aquantity=150uom=KGMaction=ADDbizStep=https://ref.gs1.org/cbv/BizStep-commissioningreadPointid=https://id.gs1.org/414/4054738000050{https://ns.example.com/epcis/}testField2=ABC
+eventType=ObjectEventeventTime=2015-03-30T08:58:56.591ZeventTimeZoneOffset=+02:00quantityListquantityElementepcClass=https://id.gs1.org/01/04054739999612/10/2015-03-30aquantity=150uom=KGMaction=ADDbizStep=https://ref.gs1.org/cbv/BizStep-commissioningreadPointid=https://id.gs1.org/414/4054738000050{https://ns.example.com/epcis/}testField2=ABC
+
diff --git a/tests/examples/epcisXmlDocHavingEventWithIgnoreFields.xml b/tests/examples/epcisXmlDocHavingEventWithIgnoreFields.xml
@@ -0,0 +1,59 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<epcis:EPCISDocument xmlns:epcis="urn:epcglobal:epcis:xsd:2" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
+                     xmlns:cbv="https://ref.gs1.org/cbv/"
+                     schemaVersion="2.0"
+                     creationDate="2023-01-11T12:00:00.000+01:00"
+                     xsi:schemaLocation="urn:epcglobal:epcis:xsd:2 EPCglobal-epcis-2_0.xsd"
+                     xmlns:repository-x="https://repository-x.example.com/"
+                     xmlns:example="https://ns.example.com/epcis/">
+    <repository-x:ignoreFields>
+        <example:testField1/>
+    </repository-x:ignoreFields>
+    <EPCISBody>
+        <EventList>
+            <ObjectEvent>
+                <eventTime>2015-03-30T10:58:56.591+02:00</eventTime>
+                <recordTime>2015-03-30T15:38:32.052+02:00</recordTime>
+                <eventTimeZoneOffset>+02:00</eventTimeZoneOffset>
+                <epcList/>
+                <action>ADD</action>
+                <bizStep>urn:epcglobal:cbv:bizstep:commissioning</bizStep>
+                <readPoint>
+                    <id>urn:epc:id:sgln:4054738.00005.0</id>
+                </readPoint>
+                <extension>
+                    <quantityList>
+                        <quantityElement>
+                            <epcClass>urn:epc:class:lgtin:4054739.099961.2015-03-30a</epcClass>
+                            <quantity>150</quantity>
+                            <uom>KGM</uom>
+                        </quantityElement>
+                    </quantityList>
+                </extension>
+                <example:testField1>123</example:testField1>
+                <example:testField2>ABC</example:testField2>
+            </ObjectEvent>
+            <ObjectEvent>
+                <eventTime>2015-03-30T10:58:56.591+02:00</eventTime>
+                <recordTime>2015-03-30T15:38:32.052+02:00</recordTime>
+                <eventTimeZoneOffset>+02:00</eventTimeZoneOffset>
+                <epcList/>
+                <action>ADD</action>
+                <bizStep>urn:epcglobal:cbv:bizstep:commissioning</bizStep>
+                <readPoint>
+                    <id>urn:epc:id:sgln:4054738.00005.0</id>
+                </readPoint>
+                <extension>
+                    <quantityList>
+                        <quantityElement>
+                            <epcClass>urn:epc:class:lgtin:4054739.099961.2015-03-30a</epcClass>
+                            <quantity>150</quantity>
+                            <uom>KGM</uom>
+                        </quantityElement>
+                    </quantityList>
+                </extension>
+                <example:testField2>ABC</example:testField2>
+            </ObjectEvent>
+        </EventList>
+    </EPCISBody>
+</epcis:EPCISDocument>
diff --git a/tests/examples/epcisXmlQueryDocHavingEventWithIgnoreFields.hashes b/tests/examples/epcisXmlQueryDocHavingEventWithIgnoreFields.hashes
@@ -0,0 +1,2 @@
+ni:///sha-256;cec19d047cf3b05d02cf719d9c35bc468e911d7b9f5d415ba67a074a104c28d2?ver=CBV2.0
+ni:///sha-256;cec19d047cf3b05d02cf719d9c35bc468e911d7b9f5d415ba67a074a104c28d2?ver=CBV2.0
diff --git a/tests/examples/epcisXmlQueryDocHavingEventWithIgnoreFields.prehashes b/tests/examples/epcisXmlQueryDocHavingEventWithIgnoreFields.prehashes
@@ -0,0 +1,3 @@
+eventType=ObjectEventeventTime=2015-03-30T08:58:56.591ZeventTimeZoneOffset=+02:00quantityListquantityElementepcClass=https://id.gs1.org/01/04054739999612/10/2015-03-30aquantity=150uom=KGMaction=ADDbizStep=https://ref.gs1.org/cbv/BizStep-commissioningreadPointid=https://id.gs1.org/414/4054738000050{https://ns.example.com/epcis/}testField2=ABC
+eventType=ObjectEventeventTime=2015-03-30T08:58:56.591ZeventTimeZoneOffset=+02:00quantityListquantityElementepcClass=https://id.gs1.org/01/04054739999612/10/2015-03-30aquantity=150uom=KGMaction=ADDbizStep=https://ref.gs1.org/cbv/BizStep-commissioningreadPointid=https://id.gs1.org/414/4054738000050{https://ns.example.com/epcis/}testField2=ABC
+
diff --git a/tests/examples/epcisXmlQueryDocHavingEventWithIgnoreFields.xml b/tests/examples/epcisXmlQueryDocHavingEventWithIgnoreFields.xml
@@ -0,0 +1,60 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<epcisq:EPCISQueryDocument xmlns:epcisq="urn:epcglobal:epcis-query:xsd:2" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
+                           xmlns:cbvmda="urn:epcglobal:cbv:mda"
+                           xmlns:sbdh="http://www.unece.org/cefact/namespaces/StandardBusinessDocumentHeader"
+                           schemaVersion="2.0"
+                           creationDate="2005-07-11T11:30:47.0Z"
+                           xmlns:repository-x="https://repository-x.example.com/"
+                           xmlns:example="https://ns.example.com/epcis/">
+    <repository-x:ignoreFields>
+        <example:testField1/>
+    </repository-x:ignoreFields>
+    <queryName>SimpleEventQuery</queryName>
+    <resultsBody>
+        <EventList>
+            <ObjectEvent>
+                <eventTime>2015-03-30T10:58:56.591+02:00</eventTime>
+                <recordTime>2015-03-30T15:38:32.052+02:00</recordTime>
+                <eventTimeZoneOffset>+02:00</eventTimeZoneOffset>
+                <epcList/>
+                <action>ADD</action>
+                <bizStep>urn:epcglobal:cbv:bizstep:commissioning</bizStep>
+                <readPoint>
+                    <id>urn:epc:id:sgln:4054738.00005.0</id>
+                </readPoint>
+                <extension>
+                    <quantityList>
+                        <quantityElement>
+                            <epcClass>urn:epc:class:lgtin:4054739.099961.2015-03-30a</epcClass>
+                            <quantity>150</quantity>
+                            <uom>KGM</uom>
+                        </quantityElement>
+                    </quantityList>
+                </extension>
+                <example:testField1>123</example:testField1>
+                <example:testField2>ABC</example:testField2>
+            </ObjectEvent>
+            <ObjectEvent>
+                <eventTime>2015-03-30T10:58:56.591+02:00</eventTime>
+                <recordTime>2015-03-30T15:38:32.052+02:00</recordTime>
+                <eventTimeZoneOffset>+02:00</eventTimeZoneOffset>
+                <epcList/>
+                <action>ADD</action>
+                <bizStep>urn:epcglobal:cbv:bizstep:commissioning</bizStep>
+                <readPoint>
+                    <id>urn:epc:id:sgln:4054738.00005.0</id>
+                </readPoint>
+                <extension>
+                    <quantityList>
+                        <quantityElement>
+                            <epcClass>urn:epc:class:lgtin:4054739.099961.2015-03-30a</epcClass>
+                            <quantity>150</quantity>
+                            <uom>KGM</uom>
+                        </quantityElement>
+                    </quantityList>
+                </extension>
+                <example:testField2>ABC</example:testField2>
+            </ObjectEvent>
+        </EventList>
+    </resultsBody>
+</epcisq:EPCISQueryDocument>